Background: To date, no single instrument has proved to be adequate for screening for Alzheimer's dementia (AD). The aim of this study was to identify a combination of instruments which were highly sensitive for screening late onset AD.

Methods: Subjects were drawn from the Vienna TransDanube Aging (VITA) study. This is an interdisciplinary, longitudinal community-based cohort study of the 21st and 22nd district of Vienna (Austria). Data refer to the cohort of 478 individuals at age 78 who took part in the first follow-up investigation of the VITA study. The psychometric instruments which were investigated were: the Ten-Point Clock Test, the Human-Figure Drawing Test, a Delayed Selective Reminding Test, Naming, the Trail Making Test-B, and Verbal Fluency. Further instruments were the Pocket Smell Test, and Subjective Memory Complaints. Data were analyzed using logistic regression analyses and cross validation.

Results: A combination of the Delayed Selective Reminding Test and Verbal Fluency was best for screening AD (R2 = 0.38, main model). An area under the ROC curve of 0.829 was reached. This model discriminated between subjects with incident AD and subjects who did not have incident AD with a sensitivity of 91% and a specificity of 56%.

Conclusion: The combination of an episodic memory test and a test of verbal fluency was an effective way of screening for AD.